Marketplace Ad Spend Before 11.11: What ROAS Actually Breaks Even?

If your marketplace dashboard reports a 3x return on ad spend, you are probably losing money. Once you strip out GST, platform commission, payment fees, the campaign discount and the cost of getting the parcel to the customer's door, a Singapore SME selling at a S$60 basket typically needs somewhere between 5x and 6x just to break even. The dashboard figure is not wrong, it is measuring a different thing from profit. Before 11.11 opens, work out your own break-even multiple for each product tier, write it down, and cap spend against it rather than against a target someone copied off a platform webinar slide.

What is your real break-even return on ad spend?

The arithmetic is short. Break-even ROAS is one divided by your contribution margin before advertising. If 20 cents of every revenue dollar survives all costs except ads, you need 5x. If 15 cents survives, you need 6.7x.

The work is in getting the contribution margin honest. Take a S$60 list price, GST inclusive, sold at a 15 per cent campaign discount:

• Customer pays S$51.00
• Less GST at 9 per cent, so your actual revenue is S$46.79
• Platform commission and payment processing, charged on the S$51 collected, say 8 per cent combined: S$4.08
• Cost of goods: S$27.00
• Pick, pack and last-mile delivery: S$5.50
• Returns and refund allowance at 4 per cent: S$1.87

What is left before a single dollar of advertising is S$8.34, or 17.8 per cent. The break-even multiple is 5.6x against your revenue, or 6.1x measured the way the dashboard measures it, against the S$51 collected. At the 3x the platform congratulates you for, that order loses roughly S$8.

Run this once per price tier, not once for the whole catalogue. A S$180 bundle with the same percentage costs carries the fulfilment charge far more comfortably than a S$22 accessory, and the accessory is usually where the ad budget quietly drains.

Why does the platform dashboard flatter the result?

Four reasons, all of them structural rather than dishonest.

It reports gross, not net. Attributed sales value is what the customer paid, including GST and before commission. Your profit and loss never sees that number.

The attribution window is generous. A seven-day click and one-day view window will credit the ad for a customer who saw it, ignored it, and bought two days later after searching your brand by name. That order was going to happen.

Fulfilment and returns are not in the model. The platform does not know your courier rate or your return rate, so it cannot deduct them.

Repeat buyers are counted as wins. Your best customers are the most likely to click your ad, because the platform's own targeting finds them cheapest. You pay to reacquire people you already own.

How do you work out your true cost per new customer?

Split attributed orders into first-time buyers and returning buyers. If 70 per cent of ad-driven orders came from customers who had bought before, then your cost per genuinely new customer is more than three times what the blended figure suggests. A S$12 blended acquisition cost becomes S$40.

That is only a problem if the new customer never comes back. Check your repeat rate over the last twelve months: what share of first-time buyers ordered again, and what did they spend? If 25 per cent return and spend another S$50, then a S$40 acquisition cost is defensible as an investment. If 5 per cent return, you are buying one-off revenue at a loss and calling it growth.

What is the honest test for whether the ads are adding orders?

Pause a defined slice of ad spend for 48 hours and watch total orders, not ad orders. Pick a handful of mid-performing product listings, switch their ads off, and compare total units sold against the same weekday pattern the week before.

If total orders hold steady while attributed orders collapse, those ads were taking credit for demand you already had. If total orders fall in step, the spend is genuinely incremental and should be protected. Do this test now, in the quiet weeks, not during 11.11 when the result will be impossible to read.

What should you cut, and what should you protect?

Cut first: broad discovery placements on products under 20 per cent contribution margin, any keyword where cost per click has risen faster than conversion since September, and spend on single-item low-value listings that cannot absorb the delivery cost.

Protect: defence of your own brand name, which is cheap and stops competitors intercepting customers already looking for you, and anything that lifts basket size rather than order count. Bundles beat single items at the same ad cost because the fulfilment charge is spread across more revenue.

Watch: the interaction between ads and stacked discounts. A platform voucher, a seller discount and free shipping applied together can push a product below its floor price while the campaign still shows a respectable ROAS, because ROAS does not know what the floor is.

What should you decide before the campaign opens?

Four numbers, agreed and written down where the person adjusting bids can see them:

1. By 22 October: the break-even ROAS for each price tier, and the target ROAS you will actually hold to, which should sit above break-even by whatever margin you need.
2. By 29 October: a daily spend ceiling and a stop-loss. If spend passes the ceiling before midday or ROAS sits under break-even for two consecutive days, campaigns pause until someone looks at them.
3. By 5 November: who is allowed to raise bids mid-campaign, by how much, and up to what total. Peak-day bid panic is the single most common way an SME turns a profitable campaign into a loss-making one in six hours.
4. Before go-live: the SKU exclusion list. Products that cannot clear break-even at any realistic ROAS should not be advertised at all, however well they sell.

None of this requires new software. It requires one spreadsheet, your actual rate card, your actual courier invoice, and a decision made in October rather than an explanation written in December.

Frequently asked questions

Is a lower ROAS ever acceptable during 11.11?
Yes, if you can name what you are buying. Running at break-even to clear ageing stock, or below it to acquire customers you have evidence will return, are both legitimate decisions. Running below break-even because the platform set the target for you is not. The test is whether you decided the number or inherited it.

Should I include my own time in the cost of running these campaigns?
For the break-even calculation, no, keep it to variable costs per order so the number stays usable. But do it separately once. If someone spends eight hours a week managing bids through November, that is real cost against a thin contribution margin, and it often argues for fewer campaigns managed properly rather than more managed badly.

My platform only shows blended ROAS across all products. What then?
Work at the campaign level instead and restructure so that each campaign holds products with similar margins. Mixing a 40 per cent margin bundle with a 12 per cent margin accessory in one campaign guarantees the blended figure tells you nothing actionable, and the bundle will be subsidising the accessory without anyone seeing it.